Overrated transistors

"anonymous"

Absolute max plate voltage = 1200.

Max plate current = 120mA.

Max plate dissipation = 14 watts.

So, max dissipation is 144 watts for about 1 second, single shot.

Or continuously, with a 10% average duty cycle.

One proviso, keep the screen volts below 300.

.... Phil

On a sunny day (Tue, 19 Jul 2011 08:31:17 +0100) it happened "anonymous" wrote in :

Use some CMOS switching transistors to switch in power resistors in parallel, ratio 1, 2, 4, 8. Now you have 15 power steps.

On a sunny day (Mon, 18 Jul 2011 18:37:28 GMT) it happened snipped-for-privacy@puntnl.niks (Nico Coesel) wrote in :

Did you scope Vce for peaks, was it perhaps oscillating? Did you scope Ic? Did you scope Ib? Did you measure temp?

(...)

We must be looking at different spec sheets. :)

On the sheet I see, the transistor is good for 1 A Ic at Vc = 53 V D.C. The transistor isn't specified for

--Winston

Hello Winston,

There is some difference in SOA for "your" datasheet and "my" datasheet. My datasheet is for the QM100DY-HK (Feb 1999). the 2A value is 53V. and then goes linearly to 1A for 100V (of course on the log scale).

Despite the slight difference in datasheets, the SOA problem remains.

With kind regards,

Wim PA3DJS

Nico,

If you don't use current balance resistors, please check for current sharing. The so called positive tempco in RDSon is for saturated use only.

If you use "older" mosfet they may behave as expected, but some newer switching mosfets have similar issues as BJT (so have poor DC SOA). In addition, you can have difference in VGSon that may result in bad current sharing.

Wim PA3DJS

Thanks for the hint. It seems MOSFETs have an area with a negative tempco. I'll add some current balancing resistors as well. I never expected the MOSFETs to share the load equally though. If I design for the junction to be 80 degrees above ambient then each device can dissipate 35W. There is still room to spare for devices to get hotter. I can also improve things a little bit if I make the heatsink 'live' so no insulation is required.

(...)

Hi there, Whimpie!

This one?

(More like 58 V but who picks nits?)

Wow! I've never seen a 'tertiary breakdown' line on an SOA chart before! (1) :)

I'm pretty sure the line ends at 1.3 A, 85 V for a reason. I suspect that Mitsubishi is saying that the area to the right is not guaranteed; don't complain if you were drawing

Extrapolating is not be a good idea here, IMVHO.

I agree. That is the main 'takeaway'.

Cheers. :)

--Winston

There is, 2 resistors!!

"Syphilis Allison"

... John

"Nico Coesel"

Like the BUZ901 etc.

.... Phil

So, I guess I should have put a smiley face at the end of my comment. I always forget to do that.

John S

It's not necessarily for you. One never knows who is reading

-- such as ignorant hobbyists like me who may take such comments too seriously. :)

Jon

That's him!

I did see this change in slope in other devices (like BUL138 or BUW1015). But that aren't darlingtons, just single chip BJT. So that was the reason for "guessing" the 100V/1A.

Agree, I normally don't like to design beyond such specs, I sleep better with some safety margin.

That fat transistor is a bit whimpy...

Wim PA3DJS

Um. Wimpie rather. (Cough)

(...)

Clearly I haven't looked at enough data sheets.

(...)

I was pleasantly shocked to hear Nico say that the transistor survived for several seconds outside the data sheet specs. So the manufacturer did have *some* safety margin. As you imply, we should not rely on it, though.

Yes, (Cough) you are right, Wimpie.

--Winston

But thats not very high voltage.

I tried it but the current sharing is less than crappy. Until 15 years ago I had a subscribtion to Elektor. I clearly recall they published several audio amplifiers with MOSFETs in parallel. I guess those can be added to the (long) list of non working Elektor circuits.

And I forgot something else important too: I want the dummy load to be a variable current sink, not a variable resistor. IOW back to BJT.